Analyzing the Trade-off between Outreach and Financial Performance for Indian Microfinance Institutions 1. Subham Kailthya 2.

Transcription

1 Analyzing the Trade-off between Outreach and Financial Performance for Indian Microfinance Institutions 1 Subham Kailthya 2 Abstract Microfinance institutions (MFIs) in India are undergoing a structural change where previously donor driven, subsidy-reliant institutions are graduating into for-profit, publicly listed organizations. This necessitates a close examination of the nature of inter-relatedness between the MFIs objectives of outreach and financial performance. A synthetic index, one for each dimension of performance outreach and financial stability - is created for each MFI using factor analysis. The factor scores computed are then used as dependent variables in a Seemingly Unrelated Regression (SUR) framework to identify the determinants of performance. The results suggest that expansion of coverage and sustainable financial performance are strongly inter-connected objectives. While operational efficiency, leverage and institutional structure affect financial performance, outreach is driven by better aligned staff incentives, reduction in operational costs and ability to manage greater risks. A strong learning by doing component is observed to exist in the microfinance industry. 1 This paper is an extension of the author s Master s thesis at the Madras School of Economics, Chennai. The author deeply thanks his supervisor Dr. Hema (hema@mse.ac.in), panel member Dr. Kavikumar (kavi@mse.ac.in) and Dr. Brinda (brinda@mse.ac.in) for their valuable comments and suggestions. 2 Project Consultant cum Coordinator at Madras School of Economics, Chennai.

2 1. Introduction Microfinance has become an important tool in providing access to basic financial services to the poor. By lending small amounts with compulsory and frequent repayments to groups and individuals, microfinance institutions (MFIs) enable the poor to raise income, build assets, and cushion themselves against external shocks (CGAP, 2004, p. 1). Its success has spread a new wave of enthusiasm amongst donors and policy makers to alleviate poverty through provision of basic financial services. This is evidenced by the fact that in 2007 more than 100 million of the world s poorest families received a microloan (Daley-Harris, 2009). In India, where 27.5 percent of the population live below the national income poverty line, microfinance can serve as a viable instrument in reversing or at least slowing down the trend of poverty. 3 Through innovations in lending mechanisms such as peer-monitored loans, weekly meetings and frequent repayments, microfinance has been able to minimize costly informational asymmetries involved in lending to the poor (Hermes and Lensink, 2007, de Aghion and Morduch, 2005) thereby ensuring self-sustainability in its operations. In general, microfinance organizations are faced with different social and operational objectives like expanding coverage, reaching out to women, increasing efficiency and earning profits to attract more capital. However, one can look at these as representing two broad objectives: expansion of outreach and stable financial performance. While providing financial access to a greater number of the poor is the core mission of MFIs, it has to be profitable in order to be sustainable (or at least to stop its reliance on subsidies). The focus of this paper is to understand explicitly the nature of inter-linkages that exist between the two MFI objectives in the Indian context. The results from this study suggest that the twin objectives of outreach and financial performance are inter-related amongst themselves. What it implies is that an expansion of coverage needs to be complemented by good financial performance which would include being efficient and productive in operations along with a strong earning potential. On the other hand, expansion of outreach would require investments in staff strength, better aligned staff compensation, reduction in operational costs and ability to manage greater risk. The 3 Poverty Estimates for , Government of India

3 microfinance industry is observed to have a significant learning by doing component where better understanding of the market translates into higher profitability. The rest of the paper is organized as follows: section 2 reviews past literature on analyzing the trade-off between outreach and financial performance; section 3 presents the data and empirical framework used in the analysis followed by section 4 which contains a detailed discussion of empirical results that identify the underlying constructs representing MFI performance and examines factors affecting such performance. Section 5 concludes. 2. Literature Review One of the detailed studies analyzing the trade-off between outreach and financial stability is by Cull and Morduch (2007). They examine outreach and profitability data for 124 institutions across 49 countries and find that the most profitable lenders are those that serve the poor to the least extent while those focusing on the poor are the most subsidy-reliant. The study does not find any evidence to suggest that raising interest rates or cost-minimization leads to greater profitability. Conning (1999) examines the contractual design issues faced by microfinance organizations that tries to maximize outreach among the poor while remaining financially sustainable. The basis of the trade-off, he suggests, is due to the costly monitoring and control system for the informationally opaque borrowers which serve as substitutes for collateral. He analyzes the data for 72 MFIs and suggests that MFIs targeting poor clients must charge higher interest rates, have higher staff cost per dollar lent, and be less leveraged. In another cross-country study covering 114 MFIs from 62 countries Hatarska and Nadolnyak (2007) examine whether regulation affects outreach and operational sustainability of MFIs and find no direct linkages. Following the approach of Luzzi and Weber (2006), outreach and financial performance of Indian MFIs are considered as multi-dimensional indices and factors affecting such indices are identified. 3. Data and empirical approach The data on microfinance institutions is obtained from the Microfinance Information Exchange (MIX), a not-for-profit private organization that aims to promote information

4 dissemination in the microfinance industry. 4 Information was collected on a broad range of variables related to financial and operational performance, age of MFIs, loan portfolio and its quality, and the coverage of MFIs for 76 Indian MFIs listed on the MIX for the year 2009 (Table 1). A list of MFIs used in the analysis is presented in Appendix I. Table 1: Description and summary statistics of key variables Variable name Variable description Mean Median Min Max cpb Cost per borrower cpl Cost per loan (USD) op_exp_assets Operating expense/ assets der Debt to equity ratio age Age (in years) avg_albpb Average loan balance per borrower (USD) pc_women Percent of women borrowers roa Return on assets (%) roe Return on equity (%) bs Borrowers per staff member yield_gp_real Yield on gross portfolio (real) par_30 Portfolio at risk > 30 days (%) Source: MIX market While it is acknowledged that there could be a potential self-reporting bias in using MIX data, other public sources containing the required variables were not available. The choice of variables was also constrained with some MFIs choosing to not report certain data fields leaving very few observations to conduct any meaningful analysis. The variable average loan balance per borrower is a proxy for depth of outreach. A smaller value of the average loan size would mean greater coverage and a higher probability of a core poor being serviced. The percent of women borrowers represents the coverage of women by MFIs. Given that getting access to finance is relatively difficult for women, providing them access to credit is a prerogative. The variables return on assets, return on equity and operating expenses to assets ratio capture the financial performance of MFIs. An increase in returns on assets and equity indicates better financial performance and hence an improvement in the profitability of the MFIs. On the other hand, since additional expenditure reduces net profits, operating expenses correlates negatively with financial performance (Figure 1). 4 Information on the MIX is available at

5 Figure 1: Pair wise correlation matrix of select variables relating to outreach and financial performance of MFIs Variables [1] [2] [3] [4] [5] Average loan balance per borrower [1] 1 Percent of women borrowers [2] (0.000) Return on assets [3] (0.790) (0.867) Return on equity [4] (0.915) (0.296) (0.000) Operational expenses to assets [5] (0.134) (0.464) (0.000) (0.000) Source: Author s own estimates Note: [a] Figures in parenthesis indicate level of significance Financial performance and outreach are broad based concepts. It is not only important to reach out to the poor but to those who are most disadvantaged in terms of access to finance such as economically active poor women. With regard to financial performance, earning returns with greater efficiency is the goal. In order to capture the multiple dimensions of outreach and financial performance a synthetic index, one for each objective, is created. By using factor analysis one is able to extract the latent variables or the constructs from a set of variables. This method exploits the variables covariance structure to form a smaller number of heterogeneous groups representing the underlying constructs or factors driving the observed correlation structure. Within each of these groups, however, the observations are largely homogeneous. Factor analysis assumes that each measured variable, say x i, can be written as the sum of a set of unobservable latent factors f j and unique (or specific) factors, ε i. It can be written as: m x = l f + εi i=1,2,.,p...(1) i ij j j= 1 or in vector notation, X( p 1) = L( p m) F( m p) + ε ( p 1)...(2) where X vector includes all observed (standardized) variables, L is the matrix of factor loadings, F is the vector of latent factors and ε constitutes the specific factors.

6 One of the problems in estimating factor loadings is that they are not uniquely determined (i.e the identification problem). Thus, the loadings * L = LT and L (3) where T (mxm) is an orthogonal matrix, so that ' ' TT T T I = = (4) both give the same representation. In such cases devising simple structures facilitates identification and interpretation of factors. The Principal Component Factor has been used to estimate the factor loadings. In general one requires the first factor to have maximal contribution to the common variance, the second to have maximal contribution to this variance subject to being uncorrelated with the first, and so on. However, in most of the cases it is possible to have a more interpretable solution by using a transformed model. This can be obtained through appropriate rotation of factors. Since our theory has evidence to justify the presence of correlated factors, oblique rotation is used (see Hetzel, 1996). Various methods exist for oblique rotation of factors and the Procrustean or promax with power 3 is used in this analysis. Once the final set of factors is obtained, a score indicating performance both financial and outreach is estimated. Factor scores are estimated as F = Z L...(5) using the Regression method where F gives the estimated factor scores, Z is the data matrix with p variables standardized to have zero mean and unit variance and L is the vector of estimated factor loadings. 4. Empirical results 4.1 Identifying the optimal number of factors The correlation between the variables (see Figure 1) indicates a significantly negative and high correlation between percent of women borrowers and average loan balance per borrower. No significant correlation is observed between percent of women borrowers and the financial performance variables return on assets and equity and operational expenses as

7 a fraction of total assets. On the other hand, the association within the financial performance variable are highly significant and has the expected signs. The next step involves choosing the optimal number of dimensions which best describe the covariance structure of the variables. One way of identifying the optimal number of factors is by using the Kaiser-Guttman or the Eigenvalue-one criteria, i.e keeping only those factors which have eigenvalues greater than 1 (Kaiser, 1960). Another method is to keep just enough factors so that the cumulative variance explained in no less than 70 per cent. (Figure 2) Figure 2: Unrotated factors using the Principal Component Factor (PCF) method Factor Eigenvalue Difference Proportion Cumulative Factor Factor Factor Factor Factor LR test: independent vs. saturated: chi2(10) = Prob>chi2 = Source: Author s own estimates The Scree plot (see Figure 3) can also be used as a visual aid in determining the number of factors to retain (Cattell, 1966). The point at which the curve bends is considered to indicate the maximum number of factors to extract. All the methods indicate the retention of two factors as the best representation of multidimensionality. Figure 3: Scree plot to identify optimal number of factors to be retained Source: Author s own estimates The un-rotated factor loadings corresponding to the variables are presented in Figure 4.

8 Figure 4: Factor loadings and unique variances (unrotated) of the retained factors Variable Factor1 Factor2 Uniqueness avg_albpb pc_women roa roe op_exp_assets Source: Author s own estimates The factors are then rotated using the method of oblique rotation to yield a better solution that is more interpretable (i.e. with a better simple structure (Thurstone, 1947)) by redistributing the variance that has been previously explained by the extracted factors. The choice of method is influenced by the assumption of inter-relatedness within the factors. The rotated factor loadings for the two factors Factor 1 and Factor 2 are presented in Figure 5. Figure 5: Rotated factor loadings and unique variances of the retained factors Variable Factor1 Factor2 Uniqueness avg_albpb pc_women roa roe op_exp_assets Source: Author s own estimates For Factor 2, the average loan balance per borrower loads negatively while percent of women borrowers loads positively. A lower average loan balance with a high percent of women borrowers would ensure that microfinance reaches to the marginalized sections of the society. The other variables do not load significantly to Factor 2. On the other hand, for Factor 1, return on assets and equity load positively while the ratio of operational expenses to assets loads negatively. Factor 2, therefore, captures the dimension of financial performance. 4.2 Determinants of MFI performance The presence of confounding factors makes it difficult to evaluate MFIs performance. In order to overcome this problem, a synthetic index, one for each dimension of performance is computed. The scoring coefficients used to estimate the factor scores for Factor 1 and Factor 2 is given in Figure 6.

9 Figure 6: Scoring coefficients obtained using regression method to estimate factor scores for Factor1 and Factor2 Variable Factor1 Factor2 avg_albpb pc_women roa roe op_exp_assets The factor scores corresponding to outreach and financial performance are used as dependent variables in two separate equations. The impact of different variables on outreach and financial performance is jointly determined in a system of equations where we assume a contemporaneously correlated error structure. Denoting the score of each MFI on Factor 1 (financial performance) as y 1 and that of Factor 2 (outreach) as y 2, the Seemingly Unrelated Regression (SUR) model estimated is: 5 y1 X1 0 β1 ε1 y = + 0 X β ε (6) where X is a vector of independent variables and β its corresponding coefficient vector. The assumptions on the error terms are E[ ε X, X ] = E εε X X ' [ 1, 2] =Ω...(7) where σ I σ I Ω= σ21i σ22i...(8) The p-value of for the Breusch-Pagan statistic rejects the hypothesis of uncorrelated errors, making OLS estimates inconsistent. A SUR model with contemporaneously correlated errors assumption is therefore used to yield parameter estimates (Figure 7). 5 See Appendix A.1 for the estimation procedure.

10 Figure 7: Coefficient estimates obtained using Seemingly Unrelated Regression (SUR) framework explaining financial performance and outreach of MFIs Financial Performance Outreach Variables [a,b] Coefficients p-value Coefficients p-value bs 0.238*** * cpl *** age 0.171** der 0.152** yield_gp_real 0.414*** NGO dummy * other dummy 1.069*** *** regulated dummy cpb *** avg_sal_gni 0.448*** par_ * Number of observations R-square Breusch-Pagan test of independence: chi2(1) = 8.749, Pr = Source: Author s own estimates Note: [a] *** p<0.01, ** p<0.05, * p<0.1 [b] All continuous variables have been standardized. Number of active borrowers per staff The number of active borrowers per staff member is an indicator of efficiency. Higher this number the more efficient is an MFI. It is highly positively significant in affecting financial performance. However, it negatively affects outreach. One possible reason is that microfinance operates in environments with little or even no financial literacy. The staff play a key role, through financial literacy programmes and other services to make the poor aware of microfinance programmes, its products and respective utility. Increase in the take-up of loans and other services are therefore observed to be directly related to the number of staff employed. Cost per loan and Cost per Borrower Cost per loan measures the average cost incurred by the MFI per unit of loan lent. As expected, a greater cost per unit of loan has a significantly negative effect in affecting financial performance of a MFI. The cost per borrower is included in the outreach equation and is significantly negative. It implies that for a MFI to expand its outreach it has to reduce its cost per borrower.

11 Age of MFI The age variable measures the number of years for which the MFI has been in existence. The results show that with an increase in age, MFIs tend to be perform better financially. This might be due to a better understanding of the market conditions and achieving a minimum portfolio size which make it profitable. However, age has no significant effect in affecting outreach. Debt-Equity Ratio The debt-equity ratio shows to what extent capital is leveraged to finance operations. Since additional funds are required to finance expansion plans which in turn would increase interest income of MFIs, an increase in the debt-equity ratio has a significant and positive effect in increasing financial performance. Real Yield on Gross Loan Portfolio This variable is an indicator of the real rate of interest charged by MFIs. While an increase in the real yield on gross loan portfolio has a significant and positive effect on the financial performance, it is not significant in affecting outreach. This might indicate that the marginal return on additional funds for the poor is high and an increase in the real rate of interest does not significantly affect outreach. Average salary to Gross National Income The results suggest that a higher salary per staff is associated with an increase in the outreach of MFIs. Microfinance is information-intensive where costly monitoring and control substitute requirements for collateral. However, monitoring itself is not free from moral hazard and therefore proper incentives is to be provided to the staff in order to maintain decent portfolio quality (Conning, 1999). Since an increase in outreach implies serving a poorer clientele, the monitoring costs including staff cost rises as well. Portfolio at Risk (less than 30 days) The portfolio at risk is the dollar value of all loans that have one or more instalments of principal due for more than 30 days. An interesting result of this analysis is that this variable significantly and positively affects outreach. Therefore, outreach can be expanded only with additional risk bearing.

12 Legal Status The legal set-up of MFIs affects outreach and financial performance. It is observed that a MFI set-up as a non-government organization (NGO) has a greater outreach compared MFIs operating as non-banking finance corporations (NBFCs). However, there is no significant difference in the financial performance of a MFI operating as a NGO compared to one operating as a NBFC. The other dummy includes banks, credit unions and other type of legal organizations. The performance of these organizations collectively with respect to outreach is poorer than MFIs operating as NBFCs. However, MFIs in the other category perform better in terms of financial performance compared to NBFCs. Regulation MFIs differ from financial institutions in that they can be either regulated or unregulated. Regulation takes the form of entry restrictions and/or prudential supervision (Hatarska and Nadolnyak, 2007). In this study regulation does not show any significant impact in affecting either financial performance or outreach. 5. Conclusion The assumption of the existence of interdependence between financial performance and outreach is intuitive. While expansion of outreach is associated with an increased efficiency of MFIs, greater risk and better alignment of staff compensation, financial performance can be improved with reductions in the cost per loan, raising the rate of interest and greater leverage of capital. The results also indicate that there is a substantial learning component involved resulting in better performance of matured MFIs.

13 Reference Armendáriz de Aghion, Beatriz and Jonathan Morduch (2005) The Economics of Microfinance, Cambridge, MA: MIT Press Catell, R.B. (1966) The scree test for the number of factors, Multivariate Behavioral Research, 1, CGAP (2004) Building inclusive financial systems. Donor guidelines on good practice in microfinance, Washington DC: Consultative Group to Assist the Poor Cull, R. Demirgüç-Kunt, A. and Morduch, J. (2007) Financial performance and outreach: a global analysis of leading microbanks, The Economic Journal, Vol. 117 (517), pp. F107 F133 Daley-Harris, S (2009): State of the Microcredit Summit Campaign Report 2009, Microcredit Summit Campaign, Washington DC. Ferro-Luzzi, Giovanni and Weber, Sylvain. (2006) Measuring the Performance of Microfinance Institutions. Available at SSRN: Greene William.H (2003), Econometric Analysis, Pearson Education Inc. Hatarska, V. and Nadolnyak, D. (2007) Do regulated microfinance institutions achieve better sustainability and outreach? Cross-country evidence, Applied Economics, 39, pp Hermes, N. and Lensink, R. (2007) The empirics of microfinance: what do we know?, The Economic Journal, Vol. 117 (517), pp. F1-F10 Hetzel, R.D. (1996) A primer on factor analysis with comments on patterns of practice and reporting. In B. Thompson (ed.), Advances in Social Science Methodology, (Vol. 4, pp ), Greenwich, CT: JAI Press Johnson Richard.A and Wichern Dean.W (2002) Applied Multivariate Statistical Analysis, Pearson Education Inc. Kaiser, H.F (1960) The application of electronic computers to factor analysis, Educational and Psychological Measurement, 20,

14 Thurstone, L.L. (1947) Multiple Factor Analysis. Chicago: University of Chicago Press Tor Jansson et al (2003), Performance Indicators for Microfinance Institutions, Microrate & IADB

15 Appendix I List of Microfinance institutions included in the study Sl No. MFI Name/Acronym Complete MFI Name 1 AML Asmitha Microfin Ltd. 2 Arohan Arohan Financial Services Ltd. 3 ASA India ASA International India Private Limited 4 Asomi Asomi 5 Bandhan Bandhan 6 BASIX Bhartiya Samruddhi Finance Limited 7 BFL BWDA Finance Ltd. 8 BISWA Bharat Integrated Social Welfare Agency 9 BJS Belgharia Janakalyan Samity 10 BSS Bharatha Swamukti Samsthe 11 BWDC Bharathi Women Development Centre 12 Cashpor MC Cashpor Microcredit 13 Chaitanya Chaitanya India Fin Credit Privated Limited 14 CReSA CRESA FINANCIAL SERVICES PRIVATE LIMITED 15 Equitas Equitas Micro Finance India 16 ESAF ESAF Microfinance and Investments (P) Ltd 17 FFSL Future Financial Services Limited 18 GFSPL Grameen Financial Services Pvt Ltd 19 GOF Growing Opportunity Finance 20 Grameen Sahara Grameen Sahara Matia Goalpara 21 GTFS Gram Tarang Financial Services Pvt. Ltd 22 GU Gram-Utthan 23 GV Grama Vidiyal 24 HiH Hand in Hand 25 ICNW Indian Cooperative Network for Women 26 India's Capital Trust Ltd Capital Trust Limited 27 Indur MACS Indur Intideepam Mutually Aided Thrift & Credit Cooperatives' Federation Limited 28 Janalakshmi Janalakshmi (formerly SAnghamithra urban program) 29 Janodaya Janodaya Public Trust 30 JFSL Jagannath Financial Services Limited (Formerly KAS Foundation) 31 KBSLAB Krishna Bhima Samruddhi Local Area Bank Limited 32 KCIPL Kaveri Credits (India) Pvt. Ltd. 33 KOPSA KOPSA 34 Kotalipara Kotalipara Development Society 35 Mahasemam Mahasemam 36 Mahashakti Mahashakti Foundation 37 MFI Microcredit Foundation of India 38 Mimo Finance Mimoza Enterprises Finance Pvt. Ltd 39 NBJK Nav Bharat Jagriti Kendra 40 NCS Nirantara Community Services 41 NEED Network of Entrepreneurship and Economic Development 42 Pushtikar Pustikar Laghu VPBSSS Ltd 43 PWMACS Payakaraopeta Women's Mutually Aided Co-operative Thrift and Credit Society 44 RASS Rashtriya Seva Samithi 45 RGVN Rashtriya Gramin Vikas Nidhi 46 RISE Rural and Urban Innovative Social Entrepresneurship 47 Saadhana Saadhana Microfin

16 Sl No. MFI Name/Acronym Complete MFI Name 48 Sahara Utsarga Sahara Utsarga Welfare Society 49 Sahayata Sahayata Micro Finance Private Limited 50 Sanchetna Sanchetna Financial Services 51 Sanghamithra Sanghamitra Rural Financial Service 52 Sarala Sarala Women Welfare Society 53 Sarvodaya Nano Finance Sarvodaya Nano Finance Limited 54 SCNL Satin Creditcare Network Limited 55 SEIL S.E. Investments Limited 56 SEWA Bank Shri Mahila Sewa Sahakari Bank Ltd 57 SEWA MACTS SEWA MACTS Federation Limited 58 SHARE SHARE Microfin Ltd. 59 Share MACTS Share MACTS 60 SKDRDP Shri Kshetra Dharmasthala Rural Development Project 61 SKS SKS Microfinance Private Limited 62 SMILE SMILE 63 SMSS Star MicroFin Service Society, formerly SYA 64 Sonata Sonata Finance Private Ltd 65 Spandana Spandana Sphoorty Financial Limited 66 SSK Sakhi Samudaya Kosh 67 SU Sahara Uttarayan 68 SVSDF Srivardan Sociodevelopment Foundation 69 Swadhaar Swadhaar FinServe Pvt. Ltd. 70 SWAWS Sharada's Women's Association for Weaker Section 71 Trident Microfinance Annapurna Financial Services Private Limited 72 UFSPL Uttrayan Financial Services Private limited 73 Ujjivan Ujjivan Financial Services Pvt. Ltd. 74 VFS Village Financial Services Private Ltd. 75 VSSU Vivekananda Sevakendra-o-Sishu Uddyan 76 WSE Welfare Services Ernakulam